11.5 Visualization techniques

Types of data visualization

Visualization techniques turn abstract math problems into something you can actually see. When you draw a diagram, sketch a graph, or map out a relationship, you're giving your brain a concrete anchor for an abstract idea. That's what makes visualization one of the most powerful problem-solving strategies in math.

Different types of visuals work for different situations. Picking the right one can be the difference between staring at a problem blankly and spotting the solution quickly.

Charts and graphs

These are the workhorses of mathematical visualization:

• Bar charts represent categorical data with rectangular bars proportional to values. Use them when you're comparing distinct groups (e.g., test scores across five classes).
• Line graphs display trends over time or continuous data by connecting data points with lines. They're your go-to for showing how something changes.
• Pie charts show composition of a whole by dividing a circle into proportional slices. They work best with a small number of categories (around 3–6).
• Scatter plots visualize relationships between two variables by plotting individual data points. They're essential for spotting correlations.
• Area charts work like line graphs but fill the space beneath the line, emphasizing the magnitude of change over time.

Maps and geospatial plots

When data has a geographic component, spatial visualizations reveal patterns that tables never could:

• Choropleth maps use color gradients to represent data values across geographic regions (think: election maps shaded by vote percentage).
• Heat maps overlay color-coded data onto geographical maps to show spatial patterns like population density or temperature.
• Cartograms distort map areas to reflect data values while keeping shapes recognizable. A country with a larger GDP might appear physically bigger.
• Bubble maps place circles of varying sizes on maps to represent data magnitudes by location.
• Flow maps illustrate movement or connections between locations using lines or arrows, such as migration or trade routes.

Network diagrams

These visualize how things connect to each other:

• Node-link diagrams represent entities as nodes and relationships as connecting lines.
• Force-directed graphs arrange nodes using simulated physical forces so that connected nodes cluster together naturally.
• Hierarchical layouts organize nodes in tree-like structures showing parent-child relationships (like an organizational chart).
• Adjacency matrices display network connections using a grid of cells, often color-coded to show connection strength.
• Sankey diagrams visualize flow and distribution of resources across a system, with the width of each band proportional to the quantity.

Infographics

Infographics combine multiple visualization types into a single cohesive presentation. They blend icons, illustrations, typography, and data-driven graphics to tell a story. Timelines, size comparisons, and illustrated data all work together to make complex information accessible at a glance.

Key visualization principles

Good visualizations follow a set of design principles that keep the focus on the math, not the decoration.

Data-to-ink ratio

Edward Tufte introduced this concept: maximize the amount of information conveyed per unit of "ink" (or pixels, on a screen). In practice, this means stripping away gridlines, unnecessary labels, and decorative elements that don't represent data. The goal is reducing cognitive load so viewers can focus on what matters. That said, you still need enough context (axis labels, a title) to avoid oversimplification.

Color theory in visualization

Color does real work in a visualization. You can use it to encode data values, distinguish categories, or highlight important information. A few guidelines:

• Harmonious schemes (complementary, analogous, triadic) make visuals easier on the eyes.
• Saturation and brightness create visual hierarchy, drawing attention to key elements.
• Accessibility matters. Roughly 8% of men have some form of color blindness, so always choose colorblind-friendly palettes.
• Cultural context can affect how colors are interpreted, so be thoughtful about your audience.

Gestalt principles

These are principles from psychology that describe how our brains organize visual information:

• Proximity: Elements placed near each other are perceived as a group.
• Similarity: Items that look alike (same color, shape, size) seem related.
• Closure: Your brain fills in gaps to complete incomplete shapes or patterns.
• Continuity: The eye naturally follows smooth paths or lines through a visualization.
• Figure-ground: You instinctively separate foreground elements from the background, which guides where you focus.

Understanding these helps you design visuals that communicate clearly without extra explanation.

Statistical visualization techniques

Statistical visualizations help you understand distributions, relationships, and patterns in data. They're essential for hypothesis testing, spotting outliers, and identifying trends.

Box plots

A box plot summarizes a dataset's distribution in five numbers:

1. The median appears as a horizontal line inside the box.
2. The box edges mark the interquartile range (IQR), spanning the 25th to 75th percentiles.
3. Whiskers extend to the minimum and maximum values within 1.5 times the IQR from the box edges.
4. Any points beyond the whiskers are flagged as potential outliers.

Box plots are especially useful for comparing distributions across multiple groups side by side.

Histograms

Histograms show the frequency distribution of continuous data. The x-axis is divided into bins (ranges), and the y-axis shows how many data points fall into each bin.

The shape of a histogram tells you a lot: is the data normally distributed? Skewed left or right? Bimodal (two peaks)? Bin width matters here. Too narrow and the graph looks noisy; too wide and you lose important detail. Cumulative histograms show a running total of frequencies, which is helpful for percentile analysis.

Scatter plots

Scatter plots map two continuous variables against each other, with each point representing one observation. The pattern of points reveals the relationship:

• Points trending upward suggest a positive correlation.
• Points trending downward suggest a negative correlation.
• A random cloud of points suggests no relationship.

Adding a trend line or regression curve makes the overall relationship clearer. Bubble scatter plots add a third variable by varying the size of each point.

Heat maps

Heat maps display data values as colors in a two-dimensional grid. Color intensity indicates magnitude or frequency. They're particularly useful for visualizing large datasets where individual values would be overwhelming, such as a correlation matrix showing relationships between dozens of variables. Patterns, clusters, and anomalies stand out quickly in a well-designed heat map.

Interactive visualization

Interactive features let you explore data dynamically rather than just looking at a static picture.

Dynamic filtering

Filters let users selectively display data based on specific criteria. Sliders, checkboxes, or dropdown menus control which data points are visible, and the visualization updates in real time. This is powerful for exploring subsets within larger datasets or comparing different segments.

Drill-down capabilities

Drill-down features provide hierarchical exploration. You start with a high-level overview, then click on elements to reveal more detailed data or subcategories. This is common in dashboards where you might click on a region to see city-level data, then click a city to see neighborhood-level data.

Real-time data updates

Some visualizations incorporate live data feeds that refresh automatically. These use technologies like websockets or API calls to fetch current data. Real-time updates are essential for monitoring rapidly changing information (stock prices, sensor readings) and often include animation to highlight changes as they occur.

Tools for data visualization

Different tools suit different skill levels and needs.

Spreadsheet software

• Microsoft Excel and Google Sheets both offer built-in charting for quick visualizations. Google Sheets adds collaborative, cloud-based features.
• Pivot tables enable dynamic data summarization and visualization.
• Conditional formatting applies color scales or data bars directly in cells.
• Spreadsheet formulas can prepare data for more complex visualizations elsewhere.

Programming libraries

• Matplotlib (Python) provides a MATLAB-like interface for static plots.
• Seaborn (Python) builds on Matplotlib with better aesthetics for statistical visualization.
• D3.js enables custom, interactive web-based visualizations using JavaScript.
• ggplot2 (R) implements the Grammar of Graphics for flexible, layered plot construction.
• Plotly supports interactive, publication-quality graphs across multiple languages.

Specialized visualization software

• Tableau offers a drag-and-drop interface for interactive dashboards.
• Power BI integrates with the Microsoft ecosystem for business intelligence.
• Flourish provides templates for animated and interactive charts.
• Datawrapper focuses on embeddable charts for online publications.
• RAWGraphs creates complex charts from spreadsheet data without coding.

Choosing appropriate visualizations

Picking the right visualization depends on three things: your data type, your audience, and your purpose.

Data types vs. visualization methods

• Categorical data → bar charts, pie charts, or treemaps for comparing groups
• Time series data → line charts or area charts to show trends
• Hierarchical data → treemaps or sunburst diagrams
• Geospatial data → maps or cartograms for spatial relationships
• Multivariate data → scatter plots or parallel coordinates for complex relationships

Audience considerations

A technical audience can handle detailed scatter plots and box plots. A general audience will get more from bar charts and line graphs. Always consider accessibility (color blindness, screen readers) and adapt complexity to your audience's data literacy. Cultural context also influences how people interpret symbols, colors, and chart types.

Purpose of visualization

• Exploratory visualizations support your own analysis and pattern discovery.
• Explanatory visualizations communicate specific findings to others.
• Descriptive visualizations summarize key characteristics of a dataset.
• Predictive visualizations illustrate forecasts or future scenarios.
• Prescriptive visualizations guide decision-making by showing optimal choices.

Common visualization pitfalls

Knowing what not to do is just as important as knowing what to do.

Misleading scales

• A truncated y-axis (one that doesn't start at zero) exaggerates differences between data points. A bar chart showing values of 98, 99, and 100 looks dramatic if the y-axis starts at 97.
• Inconsistent scales across multiple charts make accurate comparison impossible.
• Logarithmic scales can misrepresent data if they aren't clearly labeled.
• Dual y-axes risk implying false correlations when the two scales aren't carefully chosen.
• Area representations (circles, squares) mislead if they aren't scaled proportionally. Doubling a circle's radius quadruples its area, which exaggerates the visual difference.

Overcomplication

• 3D effects almost always obscure data rather than clarify it.
• Too many colors or patterns create visual clutter that distracts from the actual information.
• Cramming too many variables into one visualization overwhelms the viewer.
• Unnecessary decimal places (reporting 3.141592653 when 3.14 would do) imply false precision.

Inappropriate chart types

Some common mismatches to watch for:

• Pie charts for comparing many categories or showing tiny differences between slices
• Line graphs for unordered categorical data, which falsely implies a trend
• Bar charts with a baseline not at zero, which exaggerates differences
• Scatter plots for categorical data, which suggests a continuous relationship that doesn't exist
• Heat maps for sparse data, which overemphasizes isolated points

Storytelling with data

The most effective visualizations don't just display data. They tell a story that guides the viewer toward understanding.

Narrative structure in visualization

A good data story follows a structure:

1. Context: Provide background information and data sources.
2. Main insight: Present your key findings through carefully chosen visuals.
3. Tension: Highlight conflicts, surprises, or unexpected patterns.
4. Resolution: Explain the patterns with additional data or analysis.
5. Conclusion: Summarize key takeaways and their implications.

Annotations and context

Annotations turn a chart into a communication tool. Text labels highlight specific data points. Callout boxes provide additional context. Arrows guide the viewer's eye to important features. Footnotes and source citations establish credibility. Color or shape annotations group related elements together.

Sequencing information

Organize your visualizations in a logical flow that builds complexity gradually. Progressive disclosure reveals data in stages so viewers aren't overwhelmed all at once. Mixing static and animated elements can show change over time effectively. Throughout a sequence, maintain consistent visual language (same colors, same scales) so viewers don't have to re-orient themselves with each new chart.

Ethical considerations

Visualizations carry responsibility. A misleading chart can distort understanding just as effectively as a false statement.

Data integrity

• Use reliable, verified data sources and cite them properly.
• Avoid cherry-picking data that supports a predetermined conclusion while ignoring contradictory evidence.
• Represent uncertainty and margins of error when they're relevant.
• Keep data in its original context to prevent misrepresentation.

Bias in visualization

Bias can creep in at every stage, from data collection to color choice. Emotionally charged colors or imagery can sway interpretation. Showing only favorable portions of data distorts the picture. The best practice is to present the full range of data, acknowledge limitations, and consider how different audiences might interpret your choices.

Accessibility in design

Accessible design ensures your visualizations reach everyone:

• Provide alt text descriptions for screen readers.
• Use colorblind-friendly palettes and supplement color with patterns or textures.
• Design for different devices and screen sizes.
• Consider data sonification (representing data as sound) or tactile representations for users with visual impairments.